We propose to examine responses of human high-level visual cortex to individual images and image features, using new high-resolution functional MRI (fMRI) imaging methods. Our goal is to identify data driven categories of images and image features that optimally activate sub-regions of cortex. This will reveal the functional organization of higher-level cotex on a fine scale, and inform how these regions represent objects. This in turn will provide critical constraints on theories of object recognition, and have relevance to the wider issue of how the brain represents the external world. We will approach this in two Specific Aims. First, we will conduct an event-related fMRI study to examine neural responses to individual object images at high resolution. We will analyze whether the optimal images to activate a given region naturally group by category. Second, we will perform psychophysics in which only parts of these images are revealed, and calculate the relative informativeness of each location on the image to identify candidate image features. We will then conduct fMRI experiments to examine whether parts of the response to an individual image correspond to the presence of a particular image feature.